Hardware for AMD Reset Bug Fixes

Hi All,

As many of you are already aware I have been working to fix the reset bug for AMD Vega and Navi generation GPUs and so far the community support to obtain hardware for development on these platforms has been outstanding.

Over the past few months, and more so the last week I have had quite a number of people asking if I could also work on the AMD Radeon VII. Unfortunately as each of these GPUs is a different architecture the reset sequence is different per GPU, and as such I require the physical card on hand for development and testing.

While AMD have provided me with some people to guide me, the information provided is still very limited due to NDAs that AMD staff are under and as such getting things working properly requires a ton of experimentation per GPU (As evidenced by the partially functional Navi reset).

Since the Radeon VII is no longer available from my usual suppliers here in Australia I have contacted PLE Computers again asking if they are able to source one for me.

I have also asked for some older generation GPUs if they have any floating around such as a Fiji and Polaris based GPU so that I can see what I can do about these GPUs while I am at it. However please note as these are older GPUs my focus will be on the current generation GPUs and there are no promises.

Once I get some pricing (if PLE can source the parts) I will setup another GoFundMe campaign to cover the total cost and post the details here.

An Australian retailer with stock has been located, please see the below funding link :slight_smile:

12 Likes

Just a heads up that with my 5700XT and the entire Adrenalin driver run so farā€“am on 19.12.1ā€“Iā€™ve not seen the ā€œreset bugā€ even once. This is assuming that by ā€˜reset bugā€™ you mean the ā€œvideo driver has resetā€ OS messageā€“that was something I saw years ago but havenā€™t seen in quite awhile. If you mean something else, I can only say that itā€™s been a problem-free experience so far, save for the occasional minor bug manifesting sporadically, as all GPUs do from time to time. (as nothing made by humans is ā€œbug freeā€ā€¦;)) My hardware, if this might help you:

Windows10x64 v1909, build 18363.476
Latest Adrenalinsā€“19.12.1
AMD 50th Anniversary 5700XT
Aorus Master x570 UEFI bios F10
Corsair HX-850 PSU (72a x1 12v rail)
AMD Ryzen 5 3600X @ 3.8Ghz & boost
RAM 16GB 2x8GB, Patriot Viper Elite PV416G320C6K @3200mhz 16 16 16 36 1T @ 3733Mhz 18-19-19-19-38-68 1T
LG MultiDrive DVD writer SATA
Boot: Samsung 960 EVO NVMe 250GB (UEFI, Secure Boot partition)
2x 1TB WD Blue 7200 rpm S3
ST2000DM S3 2TB
ST4000DM004 S3 4TB
BenQ EW-3270U 4k, HDR10 monitor

Sorry no, this is not the reset we are talking about here. This is a reset of the GPU between boots of a Virtual Machine that uses the GPU directly.

NewEgg [USA] has the XFX Vega VII for $529 USD at the moment. Not sure if this helps.

https://www.newegg.com/xfx-radeon-vii-rx-vegma3fd6/p/N82E16814150820?Description=vega%20vii&cm_re=vega_vii--14-150-820--Product

Iā€™m in the process of building a system with the following, and already have GPU passthru working [Slot #1] via UnRAID on Catalina and Windows 10 VMs, so can post any info you need / want.

ASRock X570 Taichi
Ryzen 9 3950X
Radeon Vega VII

There are several Radeon VIIs on ebay as well if you donā€™t mind a used one.

Unfortunatly NewEgg and most other US online stores like Amazon will not ship to Australia.

I donā€™t mind a used one but since the funding is coming from the community if I purchase a used one that turns out to be faulty, thatā€™s the community funding wasted.

Do you need physical access? I can remotely share my machine with you. I can also purchase one and send you it?

Welcome @Borillion

Yes I do need physical access as the work involves hundreds of cold restarts of the system, and in some cases a complete power removal from the system to recover it.

That is a very kind offer but not the first I have received. The idea is to still setup a GoFundMe to cover the costs for whoever is providing the GPU. Once I know more I will update this thread with the details of what has been decided.

I have found an Australian retailer that still has new stock of Radeon 7s and as such I have setup a GoFundMe to raise the required funds.

1 Like

I just contributed a little over 15% to the fund. Thanks for taking the time to look into this! :beers:

5 Likes

Iā€™ve added 50% to that total. Letā€™s get this doneā€¦ we just need a bit more to solve this for everyone! :grinning:

5 Likes

Just added merely 5 dollars, hope it goes well!

4 Likes

Iā€™ve added $100 to the pool. Thanks everyone for helping make this happen :raised_hands:

3 Likes

You guys truly are all amazing!

It is no less appreciated even still, thank you. Everyoneā€™s financial position is different and for some $5 may be the difference between eating dinner or not. Thank you!

7 Likes

Hi!

Great!

I just added my $50 on gofundme.

Thanks for all your help!

4 Likes

Soooo close! Just $103 to go :smiley:

3 Likes

@gnif I figure since I need this and will benefit greatly by it working sooner rather than later, Iā€™ve upped my donation to get this to the finish line. Thanks for your help and being willing to do this.

Iā€™d be interested to learn more about the process of patching / fixing this card. Would you be willing to do a write up or video to explain some of it and how we can learn more?

6 Likes

Thank you all for your generous support, I have ordered the GPU and it is expected to arrive late this month, however postage might be delayed due to christmas combined with the massive fires in Australia at current.

I will see what I can do but honestly itā€™s a very boring process of reverse engineering what amdgpu does, and trying to determine what parts of the SOC IP need poking in what way to perform a reliable reset. Hours & hours of head scratching and failures until it works.

I finally got a nice camera (Sony A7Rii) that can record 4k, so I will set it up and try to document the process if I can.

5 Likes

I was wondering if PCI Express Interposers would help you to debug ?
Often, last generationā€™s or lesser width(x4/x8) PCI Expressā€™s Interposers are found on ebay at a relatively cheap priceā€¦

An interposer would be invaluable, but relatively cheap is still thousands of $ for even last gen.